Date   : Mon, 13 Feb 2006 09:48:16 -0000
From   : "David Harper" <dl.harper@...>
Subject: Re: MOS 3.20 v 3.50

"David Hunt" <dm.hunt@...> wrote:

> I did some performance tests using a bit of BBC Basic;
>
> Master128/Turbo/Z80 2nd proc/Cambridge Co-Processor/Laptop all BBC Basic.
>
> Simple trig. function into floating point variable
> --------------------------------------------------
>
> 10 TIME=0
> 20 FOR X%=0 TO 5000
> 30 F=SIN(X%)
> 40 NEXT
> 50 PRINT TIME/100;" seconds."
>
> 65C102@... 65C102@...
> 68.01s 33.17s
>
> Z80A@... 32016@...
> 38.21s 24.89s
>
> P3/2.26GHz
> 0.0041s
>
> Insert value into array
> -----------------------
>
> 10 TIME=0
> 20 DIM Y%(5000)
> 30 FOR X%=0 TO 5000
> 40 Y%(X%)=X%
> 50 NEXT
> 60 PRINT TIME/100;" seconds."
>
> 65C102@... 65C102@...
> 4.86s 2.36s
>
> Z80A@... 32016@...
> 6.86s 4.12s
>
> P3/2.26GHz
> 0.0016s

For completeness, in case anyone is interested, I just tried these out on 
the Master 512 (i.e. 80186 co-pro running at 10MHz). Results are:

    1. (trig. function test): 14.85 sec
    2. (array test): 4.10 sec

Fascinating how this is quite a bit faster than the 8-bit chips for the 
trig. test, but for the array test it is hardly faster at all, and indeed 
slower than the Turbo. Must be something to do with the coding.

David Harper